Rationalités correctives et intelligence artificielle assistée : les doubles contraintes des humanités numériques

This article addresses the complexity of the interaction between Artificial Intelligence (AI) and human rationalities, with a focus on the challenges posed by digital humanities. Through an exploratory study, two perspectives are articulated: on one hand, the critique of the idea that AI could faithfully imitate or replicate human rationalities, and on the other hand, an examination of the limitations and possibilities of artificial intelligence as an autonomous system. The analysis draws on extended interactions with ChatGPT, aiming to test AI’s capabilities within pedagogical and epistemic contexts. The article proposes a reconceptualization of the relationships between games, calculations, and possible worlds, highlighting the implications for different semiospheres (cultural/artificial) and standards of rationality. Finally, the author discusses the potential roles of AI as a catalyst for enunciation, a mediator in distributed cognition, and an agent participating in collective intelligence, while underscoring the intrinsic limitations of AI’s reflexive and cognitive capacities.